RE: MustUnderstand

["Klotz, Leigh" <Leigh.Klotz@xxxxxxxxx>]

May I suggest examining the W3C XForms use of mustUnderstand?
You put in a mustUnderstand attribute on the elements that are required,
rather than on the namespace declarations.
You can just put it on the first such element, since if it fails for the
first one it won't get to the rest.

See the W3C rec here:
http://www.w3.org/TR/xforms/slice3.html#module-mustUnderstand 
And Micah Dubinko's book description here:
http://xformsinstitute.com/essentials/browse/ch11.php

-----Original Message-----
From: owner-atom-protocol@xxxxxxxxxxxx
[mailto:owner-atom-protocol@xxxxxxxxxxxx] On Behalf Of Tim Bray
Sent: Thursday, June 23, 2005 8:37 AM
To: Dare Obasanjo
Cc: Joe Gregorio; James M Snell; Atom-Protocol
Subject: MustUnderstand


On Jun 23, 2005, at 7:55 AM, Dare Obasanjo wrote:

> +1 on a mustUnderstand construct
>
> It was rejected for the syndication format. A mustUnderstand construct

> is hard to justify for a syndication format but is quite necessary for

> a publishing protocol.

During the discussion of MustUnderstand for the format, we eventually
boiled it down to the minimum-imaginable approach, namely something
like, required to be the first element:

<mustUnderstand>
  <ns>http://foo</ns>
  <ns>http://bar</ns>
  ...
  </mustUnderstand>

With the semantic being "If you don't know how to process any of these
namespaces, stop here".

This was appealing in that it avoided the kind of gyrations people have
to do in dealing with SOAP headers, allowed for Fast Fail, and was real
easy to understand.

Is something more sophisticated and fine-grained required for protocol,
or can we get away with the minimum possible?  -Tim